A Unique Chicken B-creatine Kinase Gene Gives Rise to Two B-creatine Kinase Isoproteins with Distinct N Termini by Alternative Splicing *

In the chicken, a high degree of heterogeneity at the protein level has been reported for the creatine kinase-type B (B-CK). Here we show that the two B-CK isoproteins, Baand Bb-CK, are encoded by two mRNAs, which are derived from a single copy gene by a stochastic alternative splicing mechanism. The transcription of the single hnRNA is directed by a complex promoter region containing a stretch of sequences which is highly conserved among all the B-CK genes known to date. This stretch encompasses a putative binding site for the TA-rich DNA-binding protein (Hobson, G. M., Mitchell, M. T., Molloy, G. R., and Pearson, M. L. (1988) Nucleic Acids Res. 16, 8925-8944) which is located in the distal part of the promoter region, while the proximal portion containing the TATA-box used in vivo is not conserved between chicken and mammals. The two isoproteins arising from this gene contain distinct N-terminal portions. According to comparative analysis, Bb-CK is the form which is homologous to the mammalian B-CKs, whereas Ba-CK shows some sequence features unique among all other vertebrate cytosolic creatine kinases characterized [...] WIRZ, T, et al. A unique chicken B-creatine kinase gene gives rise to two B-creatine kinase isoproteins with distinct N termini by alternative splicing. Journal of Biological Chemistry, 1990, vol. 265, no. 20, p. 11656-66

In the chicken, a high degree of heterogeneity at the protein level has been reported for the creatine kinasetype B (B-CK).
Here we show that the two B-CK isoproteins, Ba-and Bb-CK, are encoded by two mRNAs, which are derived from a single copy gene by a stochastic alternative splicing mechanism.
The transcription of the single hnRNA is directed by a complex promoter region containing a stretch of sequences which is highly conserved among all the B-CK genes known to date. This stretch encompasses a putative binding site for the TA-rich DNA-binding protein (Hobson, G. M The creatine kinases (EC 2.7.3.2) catalyze the reversible exchange of high energy phosphate between ADP and creatine. In vertebrates, three distinct types of subunits, each showing its characteristic developmental regulation and tissue-specific distribution, have been described: the skeletal muscle specific M-CK,' which is also expressed in mammalian heart, B-CK, which is expressed in neuronal and smooth muscle tissues, in heart, in myoblasts, as well as in many embryonic cell types, and the mitochondrial form Mi-CK, the expression of which seems to be restricted to the tissues that * This work was supported by Grant 3497-0.86 of the Swiss National Science Foundation, a predoctoral training grant of the Swiss Federal Institute of Technology (to T. W.), and a basic research grant (to J. C. P,) from the Muscular Dystrophy Association of America. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "oduertisement" in accordance with 18  also express a cytosolic form. While the cytosol$ forms, Mand B-CK, are active as dimers, the mitochondrial subunits are bound as octamers to the outer surface of the inner mitochondrial membrane (Schlegel et al., 1988a;Schnyder et al., 1988). It has been proposed that these enzymes constitute a phosphorylcreatine shuttle (Bessman and Geiger, 1981;Tombes and Shapiro, 1985;Wallimann et al., 1989), i.e. a high energy phosphate back-up and transport system, which is especially important in tissues which have to meet sudden changes in energy demand. Several cellular structures that carry out energy consuming processes have been shown to specifically bind or be functionally coupled to active forms of CK and thus have been defined as terminals of the phosphorylcreatine shuttle system. Examples include myofibrils (Schlifer and Perriard, 1988;Wallimann et al., 1983), the Ca*+ ATPases of the sarcoplasmic reticulum (Rossi et al., 1990), brain synaptic plasma membranes (Lim et al., 1983), acetylcholine receptorrich membranes (Wallimann et aZ., 1985), as well as the mitotic spindle (Cande, 1983;Koons et al., 1982) and the polysomes (Savabi et al., 1988) of certain cell types.
The regulation of CK expression was studied extensively with a special emphasis on the isoprotein switch from B-CK to M-CK during terminal differentiation of skeletal muscle tissues (Eppenberger et al., 1964;Perriard et al., 1978) and in myogenic cell cultures (Caravatti et al., 1979;Morris et al., 1972;Turner et al., 1974). This isoenzyme transition has been shown to be regulated at the transcriptional level (Kwiatkowki et al., 1985;Rosenberg et al., 1982). Furthermore, the expression of both cytosolic isoforms is modulated by hormones. While M-CK activity is responsive to somatomedins (Ewton and Florini, 1981), B-CK appears to be regulated by estrogen (Reiss and Kaye, 1981), by parathyroid hormone and prostaglandin Ez (SSmjen et al., 1984a), as well as by 24R,25-dihydroxyvitamin D (Sijmjen et al., 198413). Further heterogeneity at the protein level could be demonstrated for the cytosolic forms (Rosenberg et al., 1981) as well hs the mitochondrial type CK of chicken (Hossle et al., 1988;Schlegel et al., 198813) and human. The multiple M-CK subspecies must be due to posttranslational modifications, since unique transcripts derived from single copy genes could be found in all species examined Jaynes et al., 1986;Trask et al., 1988). In contrast, there is evidence for at least two MI-CK transcripts derived from distinct genes in chicken (Hossle et al., 1988) as well as in human (Haas et al., 1989) and for two mRNAs coding for B-CK subspecies with distinct isoelectric points, the acidic Ba-CK and the basic Bb-CK, in chicken (Perriard et al., 1987).
In chicken, two B-CK cDNA clones, H4 (Hossle et al., 1986) and 18c, have been isolated. While H4 contains the whole protein coding region for Bb-CK (Hossle et al., 1986;Perriard Distinct B-Creatine Kinases Arise by Alternative Splicing et al., 1987) 18~ lacks the C-terminal part of the coding region and differs from H4 in the 5'untranslated region and in the N-terminal portion of the coding region (see Fig. lA). In order to understand how the mRNAs coding for Ba-and Bb-CK are generated and to get sequence information about the Ba-CK protein, we decided to clone the chicken B-CK gene. ger et al., 1977)   or 46 "C (090589-40). Protected fragments were analyzed on 7 M urea 10% acrylamide sequencing gels. Gels were exposed to x-ray films at -70 'C, together with enhancing screens for 12-20 h.
Furthermore, the resulting sequence information should enable us to investigate B-CK transcription and its regulation by hormones and during myogenesis, as well as the different tasks of the two isoproteins with respect to the energy metabolism of the B-CK expressing cells.
Here we report that a single copy B-CK gene gives rise to two transcripts corresponding to the cDNA clones H4 and 18~ by a stochastic alternative splicing mechanism and that Ba-CK is encoded by the &-type transcript.  et al., 1986) at 65 "C using the conditions suggested by the manufacturer (Pall, East Hills, NY). The filters were washed four times for 30 min at 65 "C again as indicated by the supplier of the membrane. The polymerase chain reactions for the isolation of the middle part of the gene were carried out as described earlier (Saiki et al., 1988 (Melton et al., 1984) (Soldati et al., 1990).

RESULTS
Isolation of Genomic B-CK Sequences-In chicken, two distinct B-CK type cDNA clones, H4 (Hossle et aZ., 1986) and 18~ (Perriard et al., 1987), diverging in their 5' terminal portions, have been characterized (Fig. IA). H4 contains the whole protein coding sequence as well as 32 nucleotides of the 5'untranslated leader while 18~ contains 60 nucleotides of leader and only the 5'-half of the protein coding region. The middle segment of H4 is identical to the 3' proximal portion of the coding region of 18~; in addition, the 5' terminal 12 bp of the cDNA H4 are also identical to a sequence located in the 5' noncoding sequence of 18~. Sl nuclease protection analysis revealed that the 10 nucleotides at the very N terminus of H4 described earlier (Hossle et al., 1986)  used in this publication always refer to the sequence in which these 10 nucleotides are deleted. In order to isolate genomic B-CK fragments a X L47 library of chicken erythrocyte DNA was screened using the cDNA clone H4 as a probe. With this approach we hoped to identify clones carrying H4-specific as well as 18c-specific genomic sequences.
4.75 x lo6 plaques were analyzed by plaque hybridization and four independent positive clones, T8, T56, T62 (Fig. lB), and T6 were isolated. Restriction mapping and Southern blot analysis revealed that T6 and T8 contain the same B-CKspecific coding sequences corresponding to the 3' part of the gene, whereas T56 and T62 contain the 5' end of the gene. While T56 extends in the 5' direction 10 kb beyond the 5' most Hind111 fragment hybridizing to the cDNA clone H4, T8 contains 9 kb of genomic sequences downstream of the 3' end of the gene, but unfortunately the two pairs of clones do not overlap with each other. Further analysis of T6 revealed that in this clone a rearrangement during cloning must have taken place. The vector portion of T6 was heavily altered, and its propagation was affected, which made it very difficult to get usable quantities of DNA to analyze this clone. We therefore decided not to further characterize T6.
In addition, the B-CK positive clones were clearly underrepresented in the library we have utilized. A single copy gene such as the acetylcholine receptor a-subunit gene was shown to give rise to about five independent clones/million plaques screened from the same library? The B-CK clones had a frequency of one clone/l.2 million plaques, which is about six times less.
We assume that the sequences of the missing middle part of the gene are responsible for the rearrangements in the clone T6, as well as for the underrepresentation of the B-CK clones in the library. Additional screenings with fragments from either side of the gap were also unsuccessful. Therefore, we decided to utilize an alternative method for the isolation of the missing sequences, allowing us to circumvent the initial cloning steps necessary for the construction of a X library.
According to genomic Southern blot analysis (see Fig. 6), the missing part of the gene (B-CK gap = BG) was predicted to be about 3 kb in length. This prompted us to design a strategy for the amplification of this sequence by the polymerase chain reaction (PCR, see Fig. 1B). Four oligonucleotides were synthesised (BGPCRl, 2, 3, and 4). BGPCRl and 4 are derived from intron sequences flanking either side of the gap, while the primers BGPCR2 and 3 are derived from exon sequences, which are located on the missing part of the gene. Sequence analysis of the exons flanking the gap revealed that 296 bp corresponding to the middle part of the cDNA H4 (514-809), must be encoded by the missing fragment. We speculated that these cDNA sequences would be present in two exons of about 125 and 170 bp length. These assumptions were justified by the fact that all the locations of the splice junctions within the chicken gene with respect to the protein coding portion of the cDNA sequence H4 are conserved compared with those characterized within the homologous B-CK genes in rat  and human (Mariman et al., 1987). While the primer BGPCRP represents a plus strand sequence of the postulated exon in the 5' part of the gap, primer BGPCRB is designed to be a minus strand sequence of the other postulated exon.
With these four BGPCR primers, 35 cycles of amplification were carried out, using as the starting material 1 rg of high molecular weight genomic DNA from two inbred strain chickens of different White Leghorn families. We were able to amplify four distinct fragments with the primer combinations l-4, 1-3, 2-3, and 2-4 ( Fig. 1B) from both DNAs. The amplification l-4 gave rise to a 3.1-kb fragment, which is in the expected range for the whole middle part including the sequences that overlap with the X clones on either side. The length of the gap itself is 2.9 kb. The amplifications l-3, 2-3, and 2-4 yielded fragments of 2.3, 1.6, and 2.4 kb, respectively. No differences could be found between the amplification products derived from the DNAs of the two inbred strains. In order to identify these amplification products as B-CK sequences, we performed a Southern blot analysis, using the cDNA clone H4 as a probe (not shown). The four fragments gave rise to unambiguous signals, which led us to conclude that all these sequences were part of the B-CK gene. Furthermore, we isolated the amplification product l-4 and used this as a template for polymerase chain reaction assays with the primer combinations l-3,2-3, and 2-4. These amplifications gave rise to the same fragments as found when genomic DNA was used as template. Therefore, it is likely that all the fragments produced by these amplification experiments are * M. Ballivet, personal communication. Finally, the four fragments were cloned into plasmid vectors, which allowed us to sequence the two exons with their flanking regions as well as the sequences overlapping with the X clones. All sequences have been read up to five times on independent clones in order to make sure that mutations introduced by the inaccuracy of Taq polymerase could be unambiguously identified.
The two exon sequences matched perfectly to the respective sequences of the cDNA clone H4 and they are both flanked by canonical splice acceptor and donor sequences (Mount, 1982). In addition, the 5'-terminal as well as the 3'-terminal sequences of the fragment l-4, which spans the whole middle part, are identical to the sequences deduced from the flanking X clones.
Two Transcripts Corresponding to the cDNA Clones H4 and 18~ Arise by an Alternative Splicing Mechanism-Sequence analysis of the genomic fragments that hybridize to the cDNA probes ( Fig. 1B) revealed that exons coding for the cDNAs H4 and 18c are intermingled (Fig. lB, 8). In light of the fact that H4 codes for a B-CK protein (Hossle et al., 1986) and that H4 as well as 18c have been shown by Sl analysis (Perriard et al., 1987) to correspond to mRNAs expressed in B-CK-containing tissues, we concluded that these two cDNA clones correspond to two B-CK mRNAs arising from one gene by an alternative splicing mechanism. The two alternatively spliced exons 2a and 2b span the parts of the two cDNAs between the regions of sequence identity (Fig. lA, black and white boxes). The H4-specific exon 2b (Fig. 3, nucleotides 771-983) contains the cDNA sequence from nucleotide 13 to 225 and the 18c-specific exon 2a (Fig. 3, nucleotides 1380-1567) from 51 to 238 of the respective cDNA. Since the two exons contain the two putative translation start sites (Fig.  lA), the two mRNAs code for two B-CK protein species with distinct N termini (see below).
In addition to this feature, which is unique among all CK genes characterized so far, the chicken B-CK gene is about 9 kb long, which is almost three times the length of the mammalian B-CK genes. The lengths and the number of the exons/transcript, as well as the location of the splice sites in the gene with respect to the protein coding regions of the cDNA sequences, however, are almost identical to those in the mammalian genes. Only the exons 1, 2a, 2b, and 8 differ in length by some base pairs due to species-specific differences in the 5'untranslated leader and the 3' trailer sequences, respectively. All exons are flanked by canonical splice sites, and potential lariat acceptor sequences could be found in the expected locations near the splice acceptor sites.
Characterization of Exon 1 and Localization of the Transcription Initiation Site-The mRNAs corresponding to the two cDNAs were found to be coexpressed in equal amounts in all the B-CK-containing tissues tested so far (Perriard et al., 1987). We therefore adopted the working hypothesis that the two transcripts could have a common first exon and be under the control of a single promoter.
Such a putative exon 1 was found just upstream of the alternatively utilized H4-specific exon 2b. Its 3' end consists, as expected, of the 12-bp stretch present in the 5' portions of both cDNAs and in the 5' direction the 18c-derived 5'terminal sequence follows (Fig. L4, dark hatched box). This exon is followed by a canonical splice donor site, however, no splice acceptor could be detected at the location, where the homology between the genomic sequences and the cDNA 18c ended. Furthermore, about 50 bp further upstream, two promoter-like elements were found, each composed of a perfect TATA consensus sequence and a CCAAT box.
In order to test if these putative promoter elements are the ones used in viva, we performed the Sl protection experiments and primer extension analysis shown in Fig. 3. Probe A (oligonucleotide 120589-46) spans the region from the postulated exon 1 (nucleotide number 32 of the cDNA sequence 18~) up to the distal TATA-like element (-49 to +58 in Fig.  2) and is followed by 11 bp of extraneous sequence. This probe has been used in Sl protection experiments with different RNAs isolated from B-CK containing tissues such as gizzard (Fig. 3A, lane 4), whereas RNA from adult skeletal muscle served as a negative control (Fig. 3A, lane 3). With gizzard RNA four major protected fragments of 58, 57, 56, and 51 bp length were seen. The same was true for RNAs from brain and heart (not shown) while no band occurred in the leg muscle RNA. The longest fragment extends up to a G nucleotide 22 bp 3' of the proximal TATA-box.
These results were corroborated by data obtained from primer extension analysis with two different synthetic oligonucleotides, one of them being H4-specific (Fig. 3B) and the other being directed against 18c-derived sequences, encoded mainly by exon 1 (Fig. 3C). The H4-specific probe B (oligonucleotide 040788-79) hybridizes to the transcripts in a region which is encoded by the alternatively used exon 2b (771-802, Fig. 2) and extends into exon 1 by 8 nucleotides (69-76, Fig.  2) i.e. the probe covers nucleotides 5-44 with respect to the cDNA sequence. Reverse transcription of gizzard poly(A)+ RNA gave rise to two products of 107 and 101 bp in length, respectively (Fig. 3B). Again, the same results were obtained with RNAs from brain and heart. The main extension product therefore maps to a G residue 1 bp further 3' than the longest protected fragment found in the Sl analysis. Thus, the values obtained by this primer extension can be considered to be in perfect agreement with the Sl data discussed above. The same is true for the primer extension performed with probe C (oligonucleotide 280688-75), corresponding to the cDNA 18c between nucleotides 5 and 54 (positions 30-76 and 1380-1385 in Fig. 2), which gave rise to two products of 78 and 72 nucleotides in length. The fact that only two products were found in the primer extension analysis is likely to be due to the lower resolution of the gels used in this assay. No transcripts corresponding to an activity of the distal B-CK promoter element were found with either Sl analysis or primer extension experiments at any developmental stage or in any B-CK expressing tissue tested.
According to our working hypothesis, the 5' end of the H4type transcript must also contain the sequences found in the 5'-terminal part of the cDNA 18c and in exon 1 (dark hatched boxes in Figs. 1 and 3). To test this hypothesis, we performed another Sl analysis (Fig. 30). The probe which was utilized spans part of the H4-specific sequence encoded by exon 2b (771-78'7 in Fig. 2, nucleotides 13-29 of the cDNA clone), the part of exon 1 (40-67, Fig. 2), which is contained within both cDNAs, as well as that which was only found in clone 18c (nucleotides 12-48 of the cDNA) and 11 bp of extraneous sequence at its 3' end. Analysis of brain total RNA revealed protection of the 18c-type sequences together with the labeled H4 end. A protected fragment of 53 bp length resulted, which indicates that exon 1 is indeed part of both transcripts derived from the B-CK gene. We therefore conclude that the proximal promoter-like element is the one which directs transcription of one type of hnRNA, and this RNA gives rise to two mature mRNAs by a non-regulated, alternative splicing mechanism. Taken together, the data presented above suggest that the 5' most exon we have found is indeed exon 1 and that the proximal promoter-like element is the one which is active in uivo. This situation is reminiscent of that in the mammals, since there too, only the proximal of two promoter elements  Daouk et al., 1988;Mariman et al., 1987). Furthermore, a sequence motif ranging from position -49 to -83 in the chicken was found to be highly conserved among the different published B-CK promoters (Fig. 6, see "Discussion"). This evidence confirms that we have identified the B-CK gene promoter, although its activity remains to be investigated by direct promoter assays. The B-CK Gene Is a Single Copy Gene-In order to determine the B-CK gene copy number, genomic Southern blot analysis have been performed, using high molecular weight genomic DNA from erythrocytes of two chickens of distinct White Leghorn inbred strains. This would enable us to detect different allelic variants or multiple genes/haploid genome.
The 2.5-kb Hind111 fragment derived from the 3' end of the gene (Fig. 1B) was used as a probe and gave rise to only one hybridization signal/lane on Southern blots, when genomic DNA from the inbred strain 2883 digested with HindIII, TuqI and KpnI was tested (Fig. 4A). The three fragments thus obtained were 2.5, 8.9, and 10.5 kb long, respectively. These data are in perfect agreement with the restriction map shown in Fig. 1B and therefore indicate that the B-CK gene in chicken is a single copy gene. The same results were obtained with DNA from the other inbred strain 2854 (not shown).
Furthermore, the cDNA probe H4 gave rise to one strong band on the EcoRI digests and five signals on Hind111 digests of genomic DNA from both strains (Fig. 4B). According to the restriction map in Fig. lB, the signals on the EcoRI digests represent the sum of two corn&rating fragments of 11.5 and 11.7 kb, respectively. However, in the case of the Hind111 digests, DNA from strain 2883 gave rise to fragments of 3.2, 2.5, 1.7, 1.5, and 0.5 kb, which is again in perfect agreement with the restriction map shown in Fig. lB, whereas analysis of strain 2854 revealed fragments of 7.2, 2.5, 1.7, 1.5, and 0.5 kb. This restriction fragment length polymorphism seems to reflect a mutation at the Hind111 site which is located just 5' to the promoter region (Fig. lB, 8) in strain 2883 and in the X clones, since the two fragments that are flanking this Hind111 site add to 7.2 kb, which is the length of the fragment seen in strain 2854. These results provide further evidence for the idea that the B-CK gene is a single copy gene. In addition to this, it indicates, that the amplified fragment containing the middle part of the gene (Fig. 1B) is indeed part of the same locus as the fragments that have been isolated in the screening of the X library. The bands of 1.5 and 0.5 kb on the HindIII-digested DNA hybridizing to the probe H4 are in perfect agreement with the exon-carrying Hind111 fragments of the amplified segment.
The Two B-CK Transcripts Give Rise to Two Protein Subspecies, Bu-and Bb-CK, with Distinct N Termini-At the protein level, two isoproteins differing in their isoelectric points could be resolved by two-dimensional gel electrophoresis (Rosenberg et al., 1981), and these are now termed Baand Bb-CK. It has been shown by cell-free translation of in vitro generated RNA and subsequent comigration of the synthetic product with purified B-CK protein fractions on twodimensional gels, that the more basic protein species, Bb-CK, is encoded by a transcript corresponding to the cDNA H4 (Perriard et al., 1987;Soldati et al., 1990).
We assumed that the Ba-CK isoform is encoded by a 18ctype transcript since the two protein subspecies as well as the two RNAs are coexpressed in similar amounts in all B-CKpositive tissues tested so far (Perriard et al., 1987). Unfortunately, the cDNA clone 18~ lacks the C-terminal half of the protein coding sequence and, in addition, a screening of a Xgtll library for a full-length 18~ cDNA was not successful. However, the structure of the gene that encodes the two transcripts strongly suggests that the 3'-terminal parts of the two RNAs are identical, i.e. that the 3'-terminal part of the cDNA H4 (white hatched bar in Fig. 1) is also part of the 18ctype transcript. We therefore constructed a chimeric cDNA clone TW18-1, which contains the 5' end of the cDNA 18c, fused to the 3'-terminal portion of H4 at the BumHI site at nucleotide 395 with respect to the H4 sequence (Fig. IA). This construct was utilized for run-off transcription with the SP6 RNA polymerase and the resulting RNA served as a template for in vitro translation in the reticulocyte lysate system. The radioactive protein obtained by this procedure (Fig. 5d) comigrated on two-dimensional gels exactly with the acid Ba-CK spot in isolated brain B-CK (see also Fig. 5~). While the synthetic protein derived from TW18-1 was dispersed throughout the whole Ba-CK spot (Fig. 5d), a minor protein species was excluded from the area occupied by Ba-CK (Fig. 5c, asterisk). This minor species was identified as the phosphorylation product of a truncated Bb-CK species, that arises by initiation of translation at the methionine 12 codon (Soldati et al., 1990). The fact that such a highly related protein is sorted out from the Ba-CK spot, while the synthetic product derived from TWl8-1 is not, strongly supports the idea that Ba-CK and the TWlS-l-derived protein are identical. The black box represents 18c-specific sequences, the uhite one stands for H4-specific sequences, the dotted box indicates a sequence that is identical in both cDNAs and the dark hatched bo.r represents the 5'.terminal sequence from the cDNA 18~ (see Fig. lA), while the crosshatched part is not contained in any of the two cDNA clones. A, Sl analysis with probe A which encompasses the putative start site for transcription and contains 11 bp of extraneous sequence added at its 3' end (oligonucleotide 12Oj89-46). 20 pg of total RNA from leg muscle (lane 3) and gizzard (Inne 4) were hybridized at 55 "C overnight to 5 '"P-labeled probe A (4.8.10" Cpm/pM), digested with 30 units of Sl nuclease during 1 h, run on a 10% acrylamide gel under denaturing conditions and exposed to x-ray films for 16 h. Lane I, pBR322 cut with HpaII; lane 2, undigested probe. H, H4-specific primer extension. "'P-labeled probe B (primer 040588-79, 5.2'10" cpm/ pM) was hybridized to 2 pg of poly(A)' RNA from gizzard (lane I) and from leg muscle (lane 2) at 50 "C for 2 h and elongated with Maloney murine leukemia virus reverse transcriptase in the presence of nonradioactive dNTPs at 3i "C for 1 h. The products were separated on a 10% acrylamide gel under denaturing conditions and exposed to x-ray films for 18 h. C, 18c-specific primer extension.
The labeled probe C (primer 280688-75, 5' 10" cpm/pM) was hybridized to 2 pg of poly(A)+ RNA from leg muscle (lane I) and from gizzard (lane 2) at 60 "C for 2 h and otherwise processed in the same way as probe B. II, Sl analysis with the chimeric probe D containing %-derived sequences from exon 1 as well as H4-specific sequences from exon 2b. Probe D (oligonucleotide 090.589-40, 4.7. 10" cpm/pM) was hybridized overnight at 48 "C to 20 pg of total RNA from leg muscle (lane 2) and gizzard (lane 3). respectively. digested at 46 "C for 1 h with 50 units of Sl nuclease, and run on a 10% denaturing acrylamide gel, which was exposed to a x-ray film for 15 h. Marker: pBR322 cut with HpaII; lane I, undigested probe.
To further confirm that the in uiuo produced Ba-CK mRNA has the same coding capacity as the TW18-l-derived transcript, and that the two are also equal at the nucleotide level, we performed hybrid-arrested translations in the reticulocyte lysate (not shown). Poly(A)' RNA from brain, which was known to give rise to both species Ba-and Bb-CK after in vitro translation, was incubated with synthetic minus strand oligonucleotides complementary to 18c-and H4-specific sequences corresponding to the alternatively utilized exons, as well as with one primer complementary to a sequence stretch around position 950 in the part shared in the 3'-terminal portions of the cDNA clones H4 and TW18-1 (white hatched portion in Fig. 1). While inclusion into the assay of the H4and 18c-specific primers led to a specific inhibition of Bb-CK and Ba-CK synthesis, respectively, the 3'-specific primer abolished the translation of both B-CK subspecies. These results support further that the 3' portion of H4 codes for the C termini of Ba-as well as Bb-CK, and that there is no additional Ba-CK encoding transcript in brain RNA. Thus, we conclude that the chimeric cDNA clone TW18-1 very likely represents the Ba-CK mRNA sequence.
Finally, the complete pattern of protein spots on twodimensional gels derived from purified B-CK protein or from immunoprecipitated in vitro translation products of poly ( to the "'P-labeled 2.5-kb Hind111 fragment (8.3. 1OR cpm/ PM) that contains the 3' end of the gene (Fig. 1B). The blot was exposed to an x-ray film for 80 h. E, 4 pg of DNA from the strains 2854 (lanes I and 2) and 2883 (lanes 3 and 4). both cut with Hind111 (lanes I and 3) or with EcoRI (lanes 2 and 4) and transferred onto a nvlon membrane were hvbridized with ?'P-labeled cDNA of clone H4 (3.2. lo9 cpm/pM) and exposed to an x-ray film for 18 h. RNA from brain (Fig. 5~) can be reconstituted by comigration of the in vitro generated protein products derived from the synthetic H4-and TW18-1 transcripts (Fig. 5~). This indicates that no additional RNA species for other isoforms of B-CK are present in chicken.

DISCUSSION
The chicken B-CK gene is the first CK gene characterized to date which codes for two proteins. An alternative splicing mechanism gives rise to two transcripts which encode two B-CK subspecies, Ba-and Bb-CK, with distinct N termini. Sl nuclease protection experiments (not shown) indicate that the exons 2b and 2a are utilized in a mutually exclusive manner. We did not find any evidence for a regulation of the splicing process. All B-CK positive tissues tested express the two transcripts as well as the two proteins in roughly equal amounts. However, we cannot rule out the possibility that there may be a physiological state of a tissue in which one of the two B-CK subspecies is preferentially generated. It is remarkable, that the two genes for the cytosolic CK isoforms in chicken are both about three times as long as those found in mammals. The chicken B-CK gene is about 9 kb long, while the genes from rat  and human (Mariman et al., 1987) are both about 3.2 kb in length. The mammalian M-CK genes are roughly 12 kb long Jaynes et al., 1986;Trask et al., 1988) (Soldati et al., 1990). D, translation product from TW18-l-derived RNA comigrated with purified B-CK from brain. genes do not appear to be longer than their mammalian analogues.
A striking sequence feature has been identified flanking the H4specific alternatively spliced exon 2b. Two very AT-rich sequences, containing a 17 bp long homo-T and a 17-bp homo-A stretch, respectively (nucleotides 584-625 and 1167-1234 in Fig. 2), are located on either side of exon 2b, forming an inverted repeat. This motif might fold into a stem-loop structure, containing the alternatively utilized exon in the loop. A computer analysis (Zuker, 1981), in which 100 bp flanking the region of interest have been included, confirmed the postulated potential secondary structure. The free enthalpy of such a structure was calculated to be -18.6 kcal/M (Tinoco et al., 1973). It is tempting to speculate that this sequence motif, even if giving rise to a secondary structure of only "middle" stability, is somehow involved in the differential splicing mechanism.
The transcription of the two mRNAs is directed by a complex promoter region, which shows a high degree of homology to the promoter regions of the mammalian B-CK genes (Fig. 6). All the B-CK genes, including that of the chicken, contain two promoter-like elements within their 5' flanking regions. The proximal elements direct the transcription of all the B-CK RNAs whereas the distal elements have been reported to be silent. In the chicken each of these two elements consists of a perfect TATA consensus sequence and a CCAAT-box, while in the mammalian systems the respective sequences of the two elements differ considerably.
There, the proximal elements contain a "TTAA" motif which has been shown to be the active "TATA" box Daouk et al., 1988;Mariman et al., 1987)  (TARP), which selectively binds to the distal TATA box . In vitro transcription studies have shown that the distal promoter element can be activated by increasing the Mg2+ concentration, which concomitantly leads to the detachment of the TARP. This indicates that the TARP is likely to be part of the machinery which is responsible for the repression of the distal element.
A sequence comparison of the 5' flanking region of the chicken gene to those of the mammalian genes revealed that the region containing the TARP-binding site as well as the two CCAAT boxes is by far the most rigorously conserved part of the B-CK promoters.
Within this stretch, which reaches from position -48 to -83 in the chicken, only two mismatches occur between the chicken and the mammalian sequences. The variability of the more proximal sequences containing the active TATA boxes is clearly much higher. It has been shown by footprinting studies and gel retardation assays as well as by competition experiments using synthetic oligonucleotides spanning parts of the conserved region, that the TARP binding depends also on some nucleotides flanking the distal TATA box. On this basis a TARPlike molecule might also be able to discriminate between the two TATA boxes in the chicken B-CK promoter. However, a high selective pressure must have been responsible for the conservation of the TARP-binding region, indicating that the TARP (or an analogous activity in chicken) fulfills an important, as yet undefined, task in the regulation of the B-CK genes. Finally, it may be important to note that the spacings of the different putative &-active elements in the promoter sequences, which are compared with each other in Fig. 6, are very well conserved except for the potential SPl-binding sites, which differ in number, orientation, and location between the different species.
At the protein level, a comparative analysis of all the CK primary structures known to date (B-, M-, and Mi-CKs) revealed that there are several stretches of amino acid sequence identity, which are dispersed throughout the whole protein coding region (Babbitt et al., 1986;Hossle et al., 1988). These stretches constitute a "CK framework," which is believed to be important for enzymatic activity and other "general" CK-specific tasks (Hossle et al., 1988). Furthermore, some residues that are conserved in cytosolic forms (M-and B-CK uersus Mi-CK) and some which appear to be specific for certain isoforms have been identified.
The most obvious difference between the mammalian B-CKs characterized so far and the chicken B-CK isoproteins is the existence of the two subspecies Ba-and Bb-CK in chicken uersus the single B-type isoforms in the mammals (see also Figs. 7 and 8 (Babbitt et al., 1986;Hossle et al., 1988). At the position marked with the black point, an isotype-specific residue was postulated (Babbitt et al., 1986) and the residues marked with a rectangle were assumed to be specific for "cytosolic" CK forms (M-and B-CKs).
The suggested that the translation start of Ba-CK is also at the first AUG of the respective transcript, since it is part of a good consensus sequence for translation start sites (Kozak, 1986). However, this could not be verified by Edman degradation of the protein due to a blocked N terminus (Quest et al., 1990). Thus, the resulting Ba-CK protein is predicted to be only 376 amino acids long, while Bb-CK as well as all other cytosolic CKs contain 381 residues (including Met-l). In addition, the Ba-CK protein does not contain the features in the N-terminal portion which were expected to be typical for cytosolic isoforms (Fig. 7). Histidine 7 and lysine 11, which are found in all cytosolic CKs characterized so far, are replaced by alanine and asparagine, respectively, in Ba-CK.
Furthermore, the isotype-specific residue 18 which is aspartate in all B-C!Ks and glutamate in all M-CKs, is glutamate in Ba-CK. This glutamate is part of a stretch of 3 glutamate residues in Ba-CK which is collinear with the respective part of the chicken M-CK sequence; further N terminally two proline residues follow, which are also found in the respective location in all Mi-CK proteins characterized. Therefore it looks as if the N-terminal part of Ba-CK is a "mosaic-CK," containing sequences from different CK isoforms. These distinct N-terminal sequence features could be responsible for some differences in the biophysical properties as well as in the biological behavior of the two B-CK subspecies. For example, it has been shown in our laboratory that additional minor B-CK protein subspecies are produced in uiuo and can be resolved on two-dimensional gels. These minor subspecies can also be generated by in vitro translation of synthetic B-CK RNAs and were shown to be due to alternative initiation of translation (Soldati et al., 1990; see also Fig. 7). The analysis of mutant Bb-CK transcripts revealed that in Bb-CK the initiation of translation can take place at methionine 12, leucine 36, methionine 30, and methionine 70, in addition to methionine 1; the product of initiation at methionine 12 is the most prominent species generated by this mechanism. In vitro translation of synthetic Ba-CK RNA also gives rise to minor species comigrating with the products of initiation at methionine 30 and the two downstream sites (not shown). However, the product of initiation at methionine 12 is absent since no methionine is present in the N-terminal part of Ba-CK. The fact that these minor species also occur in uiuo suggests that they may play a physiological role and, therefore, the potential of the Bb-CK transcript to give rise to the product of initiation at merthionine 12 must be regarded as an important difference in coding capacity compared with the Ba-CK RNA.
In the rat (Mahadevan et al., 1984), it has been shown that B-CKs are phosphoproteins, and the same could be shown for the Bb-CK isoform in chicken.4 No conclusive data are yet available on whether Ba-CK is also phosphorylated, and, if it is, whether there are differences in the phosphorylation patterns of the two subspecies. In this respect it should be noted that Bb-CK contains 4 serines more than Ba-CK due to differences in their N-terminal parts (Fig. 7). It remains to be shown whether one or the other of these residues is involved in the phosphorylation; however, this question is of great importance, since there is good evidence that the enzymatic activity of B-CK is modulated by phosphorylation." As mentioned above, Ba-CK appears to have a blocked N terminus.4 Neither the structure nor the function of this block is known to date, but it is clear that this modification is Ba-CK-specific. Therefore, we can assume that the sequence information which is needed for this modification must also reside within the very N-terminal portion of Ba-CK. The exceptional heterogeneity of the B-CK proteins in chicken might account for the observation that the chicken expresses B-CKs, but no M-CK, in the adult heart, while the mammals coexpress B-and M-CK. Furthermore, neither Mi-CK protein nor its activity could be detected in chicken gizzard at any developmental stage in spite of the presence of B-CK. It is likely that the B-CKs therefore have to fulfill additional tasks in these cellular backgrounds, as compared with those in which further isoforms are coexpressed with the B-type protein. In addition to this, it has been shown recently that the heterodimer formation of the two B-CK isoforms is